Ilze Plavgo Multiple Overlapping Deprivation Analysis (MODA) Office of Research at Innocenti.

Ilze Plavgo

Multiple Overlapping Deprivation Analysis (MODA)

Office of Research at Innocenti

2

What is MODA?Child-

centered; with the child as unit of

analysis

Life-cycle approach;

capturing age-specific needs

Deprivation overlaps; using an integrative approach between dimensions

Combines deprivation and

monetary poverty analysis; for children

Flexible methodology; selecting context-specific indicators,

dimensions, profiling variables etc.

Child deprivation profiles; Supports

equity focus

MODA

A child-centered, multidimensional approach to measuring child poverty

•Children’s needs are multidimensional and

age specific

Builds on existing initiatives to measure child poverty and deprivation:

•Sector by sector analyses

•UNICEF’s Global Study

•Alkire and Foster methodology (also used for OPHI’s Multidimensional Poverty Index, Bhutan happiness index etc.)

•Material Deprivation, Social Exclusion (Europe)

3

From sector approach to child centered approach

Single deprivation analysis Multiple deprivation analysis

4

• Children’s needs are not homogenous across childhood

• Life-cycle approach allows to select age-specific indicators and analyse the different groups of children separately depending on their age to reflect children’s different needs in the periods of infancy, childhood, and adolescence

‘Life-cycle’ approach

5

Life-cycle stages and dimensions in CC-MODA

6

MODA is flexible, not an ‘index’: “How many children suffer a deprivation?”“How many experience multiple deprivations?” versus “x% of children are multidimensionally poor.”

Côte d’Ivoire 2012: children aged 5-17 years Congo DR 2010: children aged 5-17 years

But.. what are these specific deprivations? Are there some that occur in isolation?..

7

Deprivation overlap for each dimension in Ethiopia (2011) among children aged below five, by region

MODA supports integrative approachshowing where cross-sector cooperation and linkages can be beneficial

8

Integrated approach

Overlap of nutrition, health, and water deprivations in Ethiopia (2011) among children aged below five, by region

Which deprivations do the children suffer from simultaneously? How many and who are these children? Which sectors need a multi-sectoral, integrative approach? Are there regional differences?

9

• Assists in the identification of particularly vulnerable groups • Allows to concentrate on highly deprived groups in the society and to

create profiles which assist in determining their geographical and social position

MODA Supports Equity Agenda

.228 - .297

.297 - .389

.389 - .45

.45 - .507

Deprived in Nutrition

10

MODA allows integration of monetary poverty and deprivation analysis (MALI MICS & ELIM 2009-10)

Overlap between monetary poverty (food/non-food line) and deprivation (K=3), <5 years

11

MODA applicationCC-MODA• Internationally comparable: Uses standardized definitions for age-groups,

dimensions, indicators, thresholds, profiling variables; data: MICS and DHS• Using the Convention on the Rights of the Child (CRC) as guiding principles • Process: Analysis prepared by OoR• Results available in an interactive web-portal (www.unicef-irc.org/MODA/). Currently

31 cases uploaded. Results can all be exported as a pdf or Excel. • 55 countries to be completed by June 2014

N-MODA• Country-specific analysis; Done in collaboration with UNICEF COs and their national

counterparts• Uses country-specific indicators, dimensions, thresholds, age-groups, data etc.• Various Objectives: child poverty study; SitAn; to help institutionalize child poverty

measurement into government policies and monitoring systems; etc• SAME methodology; different decisions; different (additional) elements of alnalysis

12

Framework of the MODA methodology

13

Step-by-step Guidelines to MODAGetting started

Concepts, definitions, data choice

• Step 1: Choose concept(s) of child well-being/poverty to be analyzed • Step 2: Define deprivation and specify its dimensions, based on the chosen

definition • Optional – Define any other concepts of poverty (e.g. monetary poverty)• Step 3: Select the dataset(s) and choose the unit of analysis

Preparation for the deprivation analysis

Choice of dimensions, indicators, thresholds, and age groups

• Step 4: For each dimension, select indicator(s) and define the reference population

• Step 5: Specify indicator threshold(s) and construct the indicator(s)• Step 6: Decide upon the life-cycle stages and choose age group(s) for the

multidimensional deprivation analysis • Step 7: Decide whether to impute values on other members of the same

household• Step 8: Perform a correlation test and select the final indicators• Step 9: Decide how to weight indicators and how to aggregate them into

dimensions• Optional – Perform a scalability test and select the final dimensions• Step 10: Decide how to weight dimensions• Step 11: Decide upon the identification method of the multidimensionally

deprived children

14

Step-by-step Guidelines to MODASingle deprivation analysis

Analysis by indicator and by dimension

• Step 12: Estimate a child deprivation headcount ratio for each indicator and dimension • Step 13: Select profiling variables and construct a profile of the deprived children for

each dimension• Optional - For each indicator, construct a profile of the deprived children

Multiple overlapping deprivation analysis

Deprivation count and overlap analysis per age group

• Step 14: For each child, estimate the number of dimensions in which (s)he is deprived • Step 15: Calculate the share of children deprived in 0, 1, 2, …, d dimensions, both at the

national level and by subgroup• Step 16: Carry out a deprivation overlap analysis analyzing the overlap and non-overlap

of deprivations • Step 17: Construct a profile of the children in the deprivation overlaps

Identification of multidimensionally deprived children

• Step 18: Estimate the multidimensional child deprivation headcount ratio (H) for each age group using various cut-off points

• Step 19: Estimate the average intensity of deprivation (A) among the deprived for each age group using various cut-off points

• Step 20: Estimate the adjusted multidimensional child deprivation headcount ratio (M0) for each age group using various cut-off points

• Optional - Estimate H, A, and M0 for the entire child population• Step 21: Construct a profile of the multidimensionally deprived children, calculating H

and M0 at subnational and subgroup level• Step 22: Decompose M0 by (1) subgroup, and (2) dimension

Reference: De Neubourg, C., J. Chai, M. de Milliano, I. Plavgo, Z. Wei (2012), 'Step-by-Step Guidelines to the Multiple Overlapping Deprivation Analysis (MODA)', Working Paper 2012-10, UNICEF Office of Research, Florence.

15

Monetary poverty• Measured by household income/consumption

• Measures households’ resources, means to access the goods and services necessary for household members’ survival, development, and well-being

Deprivation• Measures whether individuals have access to the basic goods and services necessary

for their survival, development, and well-being (‘things’)

• Visualisation of the realities that the individuals are facing

• Measure of the individuals’ living conditions

• MODA-specific: measure of the realisation/fulfilment of the rights of the child

Subjective poverty• Measure of perceptions; measured by the individuals’ perceptions and opinions

Monetary poor children

Subjectively poor/unhappy

children

Deprived children

Monetary poor children

Subjectively poor/unhappy

children

Deprived children

Child poverty

Step-by-step Guidelines to MODAStep 1 - Choose concept(s) of child well-being/poverty to be analyzed

16

Step 2 - Define deprivation and specify its dimensions

Table 1: Child Well-being Dimensions according to the CRC used for CC-MODA

Step-by-step Guidelines to MODA

17

Step 3 – Select the dataset(s)

• CC-MODA: internationally comparable surveys

DHS V/VI and MICS 4 data

• N-MODA application: any dataset that fits the criteria best

Example: household budget surveys; MICS; DHS; LSMS; census

Tunisia: MICS 2013

Senegal: DHS 2011 (<5 years); ESPS 2011 (≥ 5 years)

Lao PDR: LECS 5 2012-13

Mali: MICS/ELIM 2009-10


18


Dimensions Indicators Age 0-4 yr Age 5-17 yr All children

1. NutritionInfant and young child feeding √ (0-23 m)

Weight for height (wasting) √

2. HealthImmunization √ (12-59 m)

Skilled assistance at birth √

3. EducationCompulsory school attendance √ (cs)

Primary school attainment √ (cs)

4. Information Availability of information devices √

5. WaterAccess to improved water source √

Distance to water source √

6. Sanitation Access to improved sanitation √

7. HousingOvercrowding √

Roof & floor material √

8. Protection from violence

Domestic violence √ (2-14 yrs)

Step 4 - Select indicator(s) and define the reference population

CC-MODA application:

19

Step 5 - Specify indicator threshold(s); construct the indicator(s)


Dimensions Indicators Deprived if

6. SanitationAccess to improved sanitation

Household usually uses unimproved toilet facility. WHO standards.

What kind of toilet facility do members of your household usually use?

Flush to piped sewer system

Flush to septic tank

Flush to pit latrine

Flush to somewhere else

Ventilated improved pit latrine

Pit latrine with slab

Pit latrine without slab/ open pit

Composting toilet

Bucket toilet

Hanging toilet/hanging latrine

No facility/bush/field

Source: Standard DHS/MICS household questionnaire

20

Step 5 – Specify indicator threshold(s); construct the indicator(s) CC-MODA APPLICATIONTable 3: Indicator thresholds for CC-MODA



1. Nutrition

Infant and young child feeding

Child under 6 months not exclusively breastfed; child between 6-59 months living in a household where children between 6-23 months are not provided with minimum meal frequency in the last 24 hours. WHO standards.

Weight for height (wasting)

Child's weight for height is below minus two standard deviations from the international median. WHO standards.

2. Health

ImmunizationChild aged 1-4 years has not received all 3 DPT vaccinations. WHO standards.

Skilled assistance at birth

No or an unskilled birth attendant assisted with child's birth. WHO standards.

3. Education

Compulsory school attendance

Child of compulsory school age but not attending school. Country-specific data, retrieved from UNESCO.

Primary school attainment

Child beyond primary school age with no or incomplete primary education. Country-specific data, retrieved from UNESCO.

4. Information

Availability of information devices

Family has not reported having any of the following: TV, radio, phone, mobile phone, and computer. MDG.

21



5. Water

Access to improved water source

Household's main source of drinking water is unimproved. WHO standards.

Distance to water source

Time needed to collect water (go, get water, and come back) is more than 30 minutes. WHO standards.

6. Sanitation

Access to improved sanitation

Household usually uses unimproved toilet facility. WHO standards.

7. Housing

OvercrowdingHousehold has on average more than four people per sleeping room. UN-HABITAT, adjusted for the number of sleeping rooms only.

Roof & floor materialBoth roof and floor are made of natural material, which are not considered permanent. UN-HABITAT standards.

8. Protection from violence

Domestic violence Child is living in a household where a child between 2 and 14 years experiences any type of physical abuse by parents.

Step 5 – Specify indicator threshold(s); construct the indicator(s) CC-MODA APPLICATION

22

Step 6: Define life-cycle stages and dimensions used

Life-cycle stages and dimensions used for the CC-MODA analysis


23

Step 6 - Define life-cycle stages and dimensions used:

TunisiaData: MICS 2013


24

Step 7 - Decide whether to impute values on other members of the same household

It may occur that questions of interest have answers only for part of an age group (e.g. for only one child in each household). For example questions on child discipline in the MICS are only asked about one child between 2 and 14 years per household. In such an event, the following choices can be made:

• Assign indicator values only to children who have information on the indicator• Assign indicator values to all children of the same household• Assign indicator values to some children of the same household, imputing data only on

those who belong to the same gender/(age) group

Step 8 - Perform a correlation test and select the final indicators


25

Step 9 - Decide how to weight indicators and how to aggregate them into dimensions

Aggregation methods:• Union approach;

• Intersection approach;

• Intermediate cut-off approach;

where yk is the deprivation status of a child i depending on the cut-off point K; Di is the number of deprivation each child i experiences; d is the total number of possible deprivations.

CC-MODA uses the union approach to aggregate indicators in dimensions, because each deprivation is relevant to the well-being of the child.


26

Step 9 - Decide how to weight indicators and how to aggregate them into dimensions


Deprivation indicators in water dimension Uganda (DHS 2010)

Drinking water source Total: 43.0%

Only source:29%

Distance to water Total: 29.8%

Only distance: 15.8%

Intersection: 14.0%

Union: 58.8%

27

Step 10 - Decide how to weight dimensionsWeights can be used to indicate the relative importance of different deprivations. The following weighting methods are used:• Equal weighting;• Data driven weighting (e.g. frequency based, most favorable, statistical and regression

based weighting);• Normative-based weighting.

CC-MODA does not assign any explicit weights to dimensions when counting the deprivations each child suffers from. It should be noted that implicitly equal weights are assigned. This method is chosen, because it reveals exactly which deprivations children are facing, making severity and overlap analyses feasible and transparent.


28

Step 11 - Decide upon the identification method of the multidimensionally deprived children

• For the identification of the multidimensionally deprived, the number of deprivations a child experiences is compared to a cut-off point.

• The union (0<K≤1), intersection (K=d) and intermediate cut-off approach (0<K ≤d) can be used

• MODA presents the outcomes for all possible cut-off points, thus incorporating all three methods.


29


• Analysis & Results (www.unicef-irc.org/MODA) :

• - Single deprivation analysis

• - Multiple deprivation analysis • Deprivation count per child

• Deprivation distribution

• Deprivation overlap per dimension

• Deprivation overlap per 2 and 3 dimensions

• Multidimensional deprivation rate

• Multidimensional deprivation depth/intensity

• Multidimensional deprivation index and its decomposition

http://www.unicef-irc.org/MODA

30

Analysis using other measures of child poverty:

Monetary poverty analysis

Decide upon the measurement of monetary poverty

Estimate the number of children living in financially poor families

Profile the monetary poor


31

Step-by-step Guidelines to MODASingle Deprivation analysis: Analysis by Indicator and by Dimension- Starting point for the calculations:

i1, i2…………in – each child in the sample;

j1, j2………….j6 – six dimensions included in the analysis;

I11,1, I11,2……I16,10 – ten indicators forming the six dimensions, the first digit of the subscript indicating the individual, the second indicating the dimension, and the third referring to the indicator.

32

Step-by-step Guidelines to MODASingle Deprivation analysis: Analysis by Indicator and by Dimension

- Deprivation status of each child per indicator

33


- Deprivation status of each child per dimension

34


35


- Child deprivation headcount ratio for each dimension

36

Step-by-step Guidelines to MODAStep 13 – Select profiling variables and construct a profile of the deprived children for each dimension

• Profiling aims at defining who are the children with a higher probability of being deprived in a specific dimension

• The selection of profiling variables is guided by:

• The objective of the analysis

• Data availability

• The applicability of criteria of relevance, variance, coverage, absence from bias and parsimony

Examples:-Area-Region-Ethnicity-Number of children per household-Parents’ occupation, job status-Mother’s level of education-Distance to the nearest hospital

37

Step-by-step Guidelines to MODAMultiple Deprivation analysis

Step 14 – For each child, estimate the number of dimensions in which (s)he is deprived

A child-centered multidimensional approach: counting the total number of deprivations of each child to inform about the breath and the combination of deprivations experienced by each child

38


Step 14 – For each child, estimate the number of dimensions in which (s)he is deprived

Deprivation count for each child

39


Step 15 – Deprivation count and distribution

40


Step 15 – Deprivation count and distribution

41


Deprivation count and distributionNational

Rural

Urban

Ville d'Abidjan

Centre-Est

Sud sans Abidjan

Centre-Nord

Nord-est

Centre-Ouest

Centre

Nord

Ouest

Nord-Ouest

Sud-ouest

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

11.5

4.2

25.1

23.7

20.4

14.5

18.9

7.4

10.5

14.5

8

3.7

3.9

3.2

19.2

12.4

32.1

39.4

27.1

25.4

21.5

18.8

14.1

13.6

14.7

14.3

12.2

10.5

24.3

22.9

26.8

26.9

26.2

24.8

24.1

24

25.7

20.3

24.7

22.6

20.9

24.7

23.1

29.9

10.1

8.5

16.8

18.8

21.3

25.2

26.5

28.4

24.2

28.6

30.8

24.8

16.2

22.1

5.1

1.6

8.5

13.3

9.4

17.8

21

16

21.4

20.4

23.4

23.4

5.2

7.6

0.600000000000001

0

1.1

3.2

4.8

6.3

1.7

7.3

6.3

8.9

7.9

11.7

0.600000000000001

0.8

0.1

0

0

0

0

0.4

0.4

0

0.700000000000001

1.6

0.9

1.7

Cote dĪvoire: DHS MICS 2012

0 1 2 3 4 5 6

Number of deprivations

42

Multiple Deprivation Analysis:

Overlap AnalysisStep 16 - Carry out a deprivation overlap analysis analyzing the overlap and non-overlap of deprivations- Determining which deprivations a child experiences simultaneously;- Pointing towards sectors in need for an integrative approach.

Step 17 - Construct a profile of the children in the deprivation overlaps


43

Multiple Deprivation Analysis:

Overlap AnalysisStep 16 - Carry out a deprivation overlap analysis analyzing the overlap and non-overlap of deprivations


44

Multiple Overlapping Deprivation Analysis:

Identification of the multidimensionally deprived children

Step 18 - Estimate the multidimensional child deprivation headcount ratio (H) for each age group using various cut-off points

where

H - multidimensional child deprivation headcount ratio according to cut-off point K in age-group a;qK - number of children affected by at least K deprivations in the age-group a; - total number of children in the age-group a;yK - deprivation status of a child i depending on the cut-off point K;Di - number of deprivations each child i experiences;K - cut-off point.


45


Step 19 - Estimate the average intensity of deprivation (A) among the deprived for each age group using various cut-off points

Where

A - average intensity of multidimensional deprivation according to the cut-off point K for the age-group a;- number of children affected by at least K deprivations in the age-group a;d - total number of dimensions considered per child within the relevant age-group a;cK - number of deprivations each multi-dimensionally deprived child i experiences, with cK = Di * yK .

46


Step 20 - Estimate the adjusted multidimensional child deprivation headcount ratio (M0) for each age group using various cut-off points

where

M0 - adjusted multidimensional child deprivation headcount ratio among children affected by at least K deprivations in age group a;cK - number of deprivations each multidimensionally deprived child i experiences, with cK = Di * yK.

Optional - Estimate H, A, and M0 for the entire child population

Step 21 - Construct a profile of the multidimensionally deprived children, calculating H and M0 at subnational and subgroup level

47

Step 22 - Decompose each adjusted deprivation headcount ratio M0 by (1) subgroup and (2) dimension(1) The decomposition by subgroup can be calculated using the following formula:

(2) The contribution of each dimension j to the overall deprivation level is expressed as a share of the total adjusted headcount ratio M0 and can be defined as follows:

where

Pj – contribution of dimension j to the adjusted headcount ratio M0

– total number of children i deprived in dimension j while also being deprived multidimensionally according to the cut-off point Kyj=1 if child i is deprived in dimension j, and yj=0 if child i is not deprived in dimension j yk=1 if child is multidimensionally deprived with Di≥K and yk = 0 if child is not multidimensionally deprived with D i <K d – total number of dimensions used in the analysis na – total number of children of the relevant age group a.


48

Field Overlap Analysis:

Overlap Analysis between different fields of child poverty

Study the overlap between the different fields of child poverty chosen for the analysis, and profile each overlap/non-overlap groupFor example, if two fields of poverty are considered - monetary poverty and deprivation - four groups can be profiled:

1. Neither monetary poor nor deprived;

2. Deprived only;

3. Monetary poor only; and

4. Deprived and monetary poor simultaneously.


Ilze Plavgo Multiple Overlapping Deprivation Analysis (MODA) Office of Research at Innocenti.

Documents

sector approach

child deprivation profiles

material deprivation

multidimensional approach

integrative approach

ccmoda slide

integrated approach

beneficial slide